Support for N-dimensional variables

Hi all!
I noticed that cvxpy added support for n-dimensional variables relatively recently ([v1.6.0](https://github.com/cvxpy/cvxpy/releases/tag/v1.6.0)), and I was wondering if something similar can be implemented in Convex.jl as well. There are a lot of applications where high-order tensors are pretty common (transports, energy systems, and warehousing are the ones I'm familiar with), so I think it would be a convenient addition.

I started going through the code, and I plan to begin tinkering with it myself, but I opened this issue for possible feedback on things that should require some attention. The first ones that pop to mind are:
- limiting positive semidefiniteness to matrices
- handling transposition, reshape, etc.